Vigil Guard

Protection Against Prompt Injection

Real-time detection of AI-specific threats with configurable sensitivity for each detection layer. Multiple detection engines operate in parallel to identify threats that single-layer solutions fail to detect.

Language Detection

Supports 22 languages with per-language ALLOW or BLOCK policies. Each prompt is processed using detection models and language-specific rule sets and dictionaries.

Semantic Drift Detection

Detects when conversations with AI drift beyond the configured assistant objective. Our proprietary cross-encoder model—built from the ground up for this task—classifies each prompt as On Scope, Near Scope, or Off Scope, enforcing role compliance in real time.

PII Protection

Built-in PII detectors with validation, plus the ability to define custom patterns for any sensitive data specific to your organization.

Shadow AI Visibility

A Chrome and Microsoft Edge extension monitors and protects employee interactions with AI platforms in real time. It provides full visibility into Shadow AI activity with instant threat detection directly in the browser.

AI Agent Security

Runtime control for autonomous agents: ALLOW, SANITIZE, or BLOCK in real time.

Content Moderation

Control over AI-generated content with native moderation in 8 languages (PL, EN, FR, ES, IT, PT, TR, RU) and per-category configuration.

Deployments in Isolated Environments

AI security for environments requiring full control over data and infrastructure: on-premise deployment via Docker, with no outbound connections, including air-gapped networks.

SIEM Integration

Native CEF or JSON export over TCP and TLS to Splunk, IBM QRadar, Microsoft Sentinel, Elastic SIEM, or any syslog collector.